Mark Sadowski recently discussed a study by Harald Uhlig on the effects of monetary shocks on RGDP. Uhlig didn’t find much effect. I suggested that there is a severe identification problem, and that NGDP fluctuations are the best indicator of monetary shocks. Mark replied as follows:
Well if all NGDP shocks were treated as monetary shocks then we’re assuming that AD shocks are purely monetary in origin. Obviously that would result in much larger estimates for the proportion of RGDP variation attributable to monetary policy shocks.
That also seems like too strong an assumption to me. That is, I think that it is implicit that monetary policy is ultimately responsible for the level of NGDP at any point in time, but that doesn’t necessarily mean monetary policy is solely responsible for every shock to AD.
I don’t think there is any question that the vast majority of economists would agree with Mark and not me on this issue. Let me try to explain why this stuff makes me a bit uncomfortable. Here are some possible definitions of monetary shocks:
1. Unexpected changes in the fed funds target
2. Deviations of the fed funds target from the Taylor Rule value
3. Unexpected changes in the monetary base
4. Unexpected changes in M2
And there are many more. For each proposed definition of a “monetary shock” you will get a different answer to the question; “How much impact do monetary shocks have on real output?” That makes it all seem quite arbitrary.
It might be helpful to return to the fiscal multiplier question as a point of comparison. The multiplier might be defined as the impact of federal spending shocks on RGDP, holding both private investment and S&L spending fixed. In fact, as far as I know economists tend to hold state and local spending fixed but not investment spending, which is allowed to vary for “crowding out” reasons when estimating the multiplier. This makes no sense to me. Federal authorities have no control over S&L spending. Almost everyone agrees that the multiplier should be estimated holding monetary policy fixed, but no one seems to know what that means.
I have a completely different view. I’m a pragmatist. For me “the” fiscal multiplier is what happens when the federal fiscal authorities change federal spending, and all other sectors of the economy (S&L, the Fed, private investment) respond to that action in the way they actually do respond in the real world. In other words, I want to know the counterfactual change in RGDP with or without that federal action, holding nothing constant.
Now let’s return to the puzzling problem of identifying monetary shocks. To me the only interesting question is how much more volatile is RGDP as compared to an economy where the Fed has adopted the optimal policy. That’s a pragmatic way to define the real effects of monetary policy. It’s a definition with policy implications. It tells us how much we can hope to improve things. So if I knew how much more unstable RGDP has been over the past few decades, as compared to the volatility of RGDP in a policy regime that pegs the price of NGDP futures contracts to rise at 5% per year (assuming that policy is optimal), then that seems to me to be the most useful definition of the contribution of monetary shocks to the business cycle. It’s a bit arbitrary, but every other definition I can think of is essentially 100% arbitrary.
Mark might reply that NGDP targeting cannot prevent all fluctuations in NGDP, which is true. This is another reason why we need an NGDP futures market. That market would provide a very good estimate of the amount of NGDP variation that was predictable, and hence preventable. If we had that market, and if it was not targeted (and hence was volatile) it would be the ideal monetary shock indicator to put into these VAR studies.